Business, science and entertainment applications depend upon computers to process and record data, often with large volumes of the data being stored or transferred to nonvolatile storage media, such as magnetic discs, magnetic tape cartridges, optical disk cartridges, floppy diskettes, or floptical diskettes. Typically, magnetic tape is the most economical means of storing or archiving the data. Storage technology is continually pushed to increase storage capacity and storage reliability. Improvement in data storage densities in magnetic storage media, for example, has resulted from improved medium materials, improved error correction techniques and decreased areal bit sizes. The data capacity of half-inch magnetic tape, for example, is now measured in hundreds of gigabytes on 512 or more data tracks.
The improvement in magnetic medium data storage capacity arises in large part from improvements in the magnetic head assembly used for reading and writing data on the magnetic storage medium. A major improvement in transducer technology arrived with the magnetoresistive (MR) sensor originally developed by the IBM® Corporation. Later sensors using the GMR effect were developed. AMR and GMR sensors transduce magnetic field changes to resistance changes, which are processed to provide digital signals. Data storage density can be increased because AMR and GMR sensors offer signal levels higher than those available from conventional inductive read heads for a given read sensor width and so enable smaller reader widths and thus more tracks per inch. Moreover, the sensor output signal depends only on the instantaneous magnetic field intensity in the storage medium and is independent of the magnetic field time-rate-of-change arising from relative sensor/medium velocity. In operation the magnetic storage medium, such as tape or a magnetic disk surface, is passed over the magnetic read/write (R/W) head assembly for reading data therefrom and writing data thereto.
As shown in FIG. 1A, a magnetic recording tape 100 typically includes a polymeric substrate layer 102, an underlayer 104, and a layer of magnetic particles (magnetic layer) 106. During writing, the magnetic moments of the individual particles in the magnetic layer 106 are oriented to represent data encoding. During readback, as the magnetic medium passes over an MR sensor, the readback signal from the MR sensor reflects a change in resistance of the MR sensor due to the influence of the magnetic medium thereon.
Readback signals are less likely to produce errors when the magnetic transitions on the tape 100 are sharp. For a given linear density, the thicker the magnetic layer 106, the deeper into the magnetic layer 106 the transitions are recorded. The result is transitions that are not sharp, i.e., the transitions have larger transition parameters, a, and broader pw50, and tis in turn can increase error rate. Deep-positioned transitions are also harder to overwrite. Accordingly, a very thin magnetic layer 106 has been found to provide the sharpest magnetic transitions. The layer of magnetic particles on modern tapes is approximately 0.1 micron thick or less.
Thus, in modern tapes, the substrate layer 102 (e.g., 7 microns thick) is significantly thicker than the magnetic layer 106. The underlayer 104 and magnetic layer 106 may be coextruded onto the substrate layer 102. The substrate layer 102 has a surface texture that is nonuniform, i.e., has asperities and other irregularities that provide surface roughness to enable reliable movement of the web through the coater. The underlayer 104 acts as a filler layer that smooths out the rough surface of the substrate layer 102. However, the coating process is not perfect, and the underlayer 104 will have an uneven upper surface which translates to the thin magnetic layer 106. Thus, after calendaring, which in part tends to make the top tape surface very smooth, the magnetic layer 106 will have areas that are thicker than others, e.g., at A, and/or will have an uneven surface, e.g., at B, which affects head-medium spacing at both A and B. In addition, agglomerations of magnetic particles are often found in the magnetic layer 106. FIGS. 1B-1E illustrate various types of asperities commonly found in the magnetic layer 106. The tape surface is shown relative to a magnetic head 150.
Magnetic recording systems suffer from Signal to Noise Ratio (SNR) degradation due to changes in signal readback amplitude due to magnetic layer irregularities, which themselves are the result of the particle agglomerations and magnetic coating thickness variations. In particular, the amplitude of the readback signal generally increases over agglomerations of magnetic particles as well as where the head-tape spacing is reduced. This variation in amplitude may increase error rate during readback signal processing.
During processing of the readback signal, the readback signal is amplified, and the amplified readback signal is processed by a detector that attempts to identify the locations of the magnetic transitions on the readback signal. There are two common approaches to analyze the readback signal. The first approach, peak detection, analyzes peak levels of the readback signal. A second approach, Partial Response Maximum Likelihood (PRML), compares the readback signal to amplitude bins or thresholds, and determines whether there is a transition based on the level and the timing of the level. While peak detection is not highly level dependent, PRML detection is very level dependent. PRML looks for signatures in the signal, and the signal characteristics must fall within a certain band or the transitions will be misdetected. When an agglomeration of magnetic particles passes by an MR head, the amplitude of the readback signal will increase dramatically. Similarly, where the magnetic layer protrudes outwardly towards the head, whether due to the agglomeration itself, or to underlayer thickness variations or a rough area on the substrate layer, the amplitude of the readback signal will increase due to the reduced head-tape spacing. The resultant amplitude shifts cause errors during peak detection and PRML detection. There are no prior known solutions to this problem.